You will need some standard packages in R, such as tidyr
, dplyr
, scipy
and ggplot2
. To run the SVM function, you need to have R packages e1071
and dismo
The SVM function is wrapped in SVM_model.R
script. You need to specify the feature genes you selected, training data, kernel function, gamma hyperparamter for kernel,
the proportion of validation set, and seed. It will return the svm model, prediction accuracy vector, and a average prediction accuracy across validation.
Main development steps are all included in the R script Including:
- Find top 100 differentially expressed genes
- Develpment of SVM model
- Select hyperparamters